You could perhaps put IR, temperature, whatever in the extra layers of a tiff and have it stitched based on the RGB.
I m using a tetracam camera that gives 5 bands per image. Pix4d processes them as a rig system. they give different weights to the different bands and they manage to generate 5 orthomosaics (one per band).
Give it a try! If your images have good overlap and they are not blurry, it should work!